Adam Reynolds won the “City Life” category (£300) at Hacked 2.1 by showing a practical commercial use of open data. His charting tool is already helping a local food startup find potential customers with food standards ratings.
What did you make?
I wrote a tool to enable companies providing services to the Food industry to identify businesses based on:
- their Food Hygiene Rating
- the underlying scores that feed into the rating (Confidence in Management, Hygiene, Structural)
- the type of business
- the date the rating was issued
- whether the rating had increased/decreased since the last time the rating had been issued
- and the geo location of the business
Specifically this tool is now being by the Bath based startup company I work for (www.sure-sense.com) to identify businesses that could have their rating improved by using the Sure Sense Catering Management system. It has real commercial value and is saving our sales guy huge amounts of time. The ability to maximises sale opportunity though a localised map on the location of the sales guy is also useful.
What did you use?
What were the challenges?
I had vaguely used Google maps in the past and was very familiar with DC.JS/Crossfilter. I wanted to see if I could get the map boundaries to reduce the set of businesses being used within the charts. MarkerCluster was then used to make the markers on the map look a LOT better. Of note there were big challenges with transforming the data into a useable data set (e.g. Theater Royal Lat/Long puts it in Bridgewater).
What would you do to improve your project further?
- Tie in the data with TripAdvisor, in particular the cost of dining at the business.
- Provide a companies house search to derive the size of the business.
- Automate the import of the FHR open data into the system. Bring in email notification of when a rating has changed (e.g. when a rating becomes a 0,1,2 or meets other notification criteria).
I am currently looking at doing a UK wide version of this tool but the map marker rendering may need to be examined. 100,000+ markers on a map is never nice. I’d also remove use of XML as this is a noise heavy protocol and provide a JSON API that would allow the removal of businesses with certain ratings (e.g. Ratings of 4 and 5 may not be something you care about but constitute the majority of businesses).
The project is live at: www.sure-sense.com/fhrs